Conformation of Inhibitor-Free HIV-1 Protease Derivedfrom NMR Spectroscopy in a Weakly Oriented SolutionJulien Roche,* John M. Louis,* and Ad Bax*[a]

Flexibility of the glycine-rich flaps is known to be essential forcatalytic activity of the HIV-1 protease, but their exact confor-mations at the different stages of the enzymatic pathwayremain subject to much debate. Although hundreds of crystalstructures of protease–inhibitor complexes have been solved,only about a dozen inhibitor-free protease structures havebeen reported. These latter structures reveal a large diversityof flap conformations, ranging from closed to semi-open towide open. To evaluate the average structure in solution, wemeasured residual dipolar couplings (RDCs) and comparedthese to values calculated for crystal structures representativeof the closed, semi-open, and wide-open states. The RDC dataclearly indicate that the inhibitor-free protease, on average,adopts a closed conformation in solution that is very similar tothe inhibitor-bound state. By contrast, a highly drug-resistantprotease mutant, PR20, adopts the wide-open flap conforma-tion.

The human immunodeficiency virus type 1 (HIV-1) protease(PR) is an aspartic hydrolase that functions as an obligatoryhomodimer with 99 amino acids in each subunit. Each mono-mer contains a glycine-rich flap segment (residues 44–57) thatfolds into an antiparallel b-sheet and covers the active site(Figure 1 A). The flexibility of the flaps is believed to controlthe access of substrates and inhibitors to the active site and istherefore recognized as essential for enzymatic activity.[1, 2] De-spite intense research over the past 20 years, the detailed roleof the flaps in the substrate binding mechanism remains sub-ject to much debate. Whereas most of the current mechanisticmodels propose that wide opening of the flaps gives access tothe active site,[3–5] a recent molecular dynamics (MD) simulationsuggests a translational procession of the substrate into theactive site while maintaining the flaps in closed or semi-openconformations.[6] The access pathway of substrate to the activesite of HIV-1 PR is of fundamental interest for gaining an im-proved understanding of how this class of enzymes channelsits substrates and is also particularly important for the designof high-affinity future-generation inhibitors.

Several hundred crystal structures of protease–inhibitorcomplexes have been deposited in the Protein Data Bank

(PDB), and in all of these complexes, the flaps adopt closedconformations. On the other hand, far fewer structural data areavailable in the absence of inhibitor, in part due to the prob-lems associated with autoproteolysis and decreased stability ofPR in the absence of inhibitors. These inhibitor-free structuresshow a variety of flap conformations, ranging from closed tosemi-open to wide open, in which the flap hairpin moietieshave moved apart by several angstroms compared to theclosed conformation.[7] Interestingly, differences between thedifferent subtypes of HIV have also been reported. For exam-ple, although the unbound PR, isolated from subtype A HIV,has only been crystallized in a closed conformation, for subty-pe B, six inhibitor-free structures were found in a semi-openconformation, two in a wide-open conformation, and only asingle one in the closed conformation.[8] These data cannot un-ambiguously distinguish whether this heterogeneity reflectsthe preferred flap conformations or results from crystal packingeffects.

Figure 1. X-ray structures of HIV-1 PR. A) Structure of PR with flaps in theclosed conformation (PDB ID: 3BVB), highlighting the flaps (yellow), thehinge regions (orange) and the active site (blue). B) Comparison of the dif-ferent structural models used in this study, showing the flap conformationsof PR in closed (3BVB, yellow), semi-open (1HHP, green) and wide-open(1TW7, cyan) conformations, as well as the PR20 structures used as referencemodels for the closed (3UCB, dark cyan) and wide-open (3UF3, red) flap con-formations. All models have been superimposed to yield a minimal Ca rmsdfor the core region (comprising residues 10–23, 62–73, and 87–93 of bothmonomers).

[a] Dr. J. Roche, Dr. J. M. Louis, Dr. A. BaxLaboratory of Chemical PhysicsNational Institute of Diabetes and Digestive and Kidney DiseasesNational Institutes of HealthBethesda, MD 20892 (USA)E-mail : julien.roche@nih.gov

johnl@niddk.nih.govbax@nih.gov

ChemBioChem 2015, 16, 214 – 218 � 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim214

CommunicationsDOI: 10.1002/cbic.201402585

The present study addresses the question concerning thepreferred flap orientations by using solution NMR spectroscopyto record residual dipolar couplings (RDCs) in a B-type activesite mutant PR in the absence of inhibitor. These RDCs can bemeasured by imposing a very slight deviation from therandom, isotropic distribution of macromolecules in an NMRsample and are very sensitive reporters on the time-averagedorientation of the corresponding internuclear vectors.[9, 10] RDCsfor the backbone N–H amide vectors were measured here foran inhibitor-free PR and a symmetric inhibitor DMP323-boundcomplex,[11] as well as for an unbound drug-resistant PR bear-ing 20 mutations (PR20) plus a D25N mutation at the activesite.[12] Weak alignment of the NMR samples was obtained bythe addition of squalamine, an antibiotic isolated from thedogfish shark, which was recently shown to form a lyotropicliquid crystal at very small volume fractions in water.[13] Thelack of any direct interaction between squalamine and PR wasevidenced by the complete absence of any detectable changein chemical shifts in the 1H–15N chemical shift correlation spec-trum, which is an exquisitely sensitive reporter of such interac-tions when present. Although amide RDCs alone are typicallyinsufficient to build a protein structure de novo, they are ex-ceptionally well suited to evaluate agreement between theactual state of the protein present in solution and the coordi-nates seen in different crystal structures; this is the approachtaken in our study.

Here, we chose three crystal structures as reference modelsfor representing the three different conformations of the flaps:PDB ID: 3BVB[14] for the closed conformation, 1HHP[15] as themost widely used model for the semi-open conformation, and1TW7[16] for the wide-open conformation (Figure 1 B). To evalu-ate movement of the flaps, we defined the PR core as theregion of the protein dimer for which predicted amide N–Hvector orientations, and thereby the predicted 1DNH RDCs, aremost invariant between these three structural models. For thisinvariant core region, excellent agreement between measured1DNH RDCs and best-fitted values was obtained for all three X-ray structures and the experimental data collected for un-bound and DMP323-bound PR, as well as for the inhibitor-freePR20 mutant, as reflected in Q-factors close to 20 % (reportedas Q-Core, Figure 2). This core region comprises the first andlast two b-sheets (residues 10–23 and 62–73), as well as thesole a-helix (residues 87–93). The five alignment tensor param-eters determined for each structural model from a singularvalue decomposition (SVD) fit to the core region were thenfixed and used to predict 1DNH RDCs for the flap region(Table 1). Here, we defined the flaps as extended regions thatincluded both the Thr80 loop and Gly40 loop (Figure 1 A), asthese two hinge segments are also involved in the openingand closing of the flaps. They are relatively invariant betweenthe different inhibitor-free and inhibitor-bound X-ray structures(Ca root mean square difference average of <0.4 �).

As expected for an inhibitor-bound complex, the flap RDCsmeasured for the PR in the presence of DMP323 agreed muchbetter with the coordinates of the closed X-ray structure, 3BVB,than those of the semi-open (1HHP) and wide-open (1TW7)reference structures, yielding Q-factors of 18.6, 32.3, and

75.8 %, respectively (Figure 2). Interestingly, much better agree-ment was also obtained when comparing the RDCs of theinhibitor-free PR flaps to the closed structure, 3BVB (Q-flap =

Figure 2. Fit of the NH RDCs measured for the inhibitor-free and DMP323-bound PR to the closed (A and B), semi-open (C and D) and wide-open (Eand F) structural models. The fit of the core region RDCs (residues 10–23,62–73, and 87–93) are shown in black. RDCs corresponding to the flaps (resi-dues 30–61 and 74–84) are displayed in red, with predicted values derivedby using the alignment tensor parameters of the core region. The qualityfactor, Q, reflects the rmsd between observed RDCs and those predicted bythe structure, normalized for the range of RDCs observed for each sample:Q = rms(Dcalcd�Dobs)/rms(Dobs), where rms is the root-mean-square function,and Dobs and Dcalcd refer to the observed RDC value and the RDC calculatedbased on the atomic coordinates, when using a best-fitted alignmenttensor.[17] To avoid an effect from the non-random distribution of the 15N�1Hvectors in the protein, rms(Dobs) was actually calculated as {Da

2[4 + 3R2]/5}1/2,where Da and R are the magnitude and rhombicity of the best-fitted align-ment tensor.[18] The experimental uncertainty in Dobs was very small(�0.15 Hz) but, due to the uncertainty in the exact positions of the hydro-gens, used for deriving Dcalcd ; even for a perfect X-ray structure, the lowerlimit on Q is ~0.15.

ChemBioChem 2015, 16, 214 – 218 www.chembiochem.org � 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim215

Communications

22.9 %) than for the semi-open (Q-flap = 42.9 %) and wide-open(Q-flap = 73.6 %) references (Figure 2). For both the inhibitor-free and DMP323-bound states, the hinge residues Thr80 andGlu35 displayed the largest deviations between the experimen-tal and predicted values when the flap RDCs were comparedto the wide-open structure, 1TW7 (Figure 2 E and F). In this re-spect, it is important to note that, even for the very best X-raystructures of any protein, RDC Q-factors will rarely fall belowabout 15–20 %, mostly limited by the fact that X-ray maps donot contain adequate electron density for defining the preciseposition of hydrogens, thus, these are typically added bymodel-building, assuming idealized geometry.[13, 18]

To verify that the backbone RDCs could capture a truly openflap conformation, we also measured RDCs for a protease bear-ing 20 mutations (PR20), found clinically in a patient with highprotease drug resistance.[19] PR20 exhibited a threefold higherdimer dissociation constant, a similar catalytic constant (kcat),an ~13-fold higher Km value for the hydrolysis of a syntheticsubstrate, and more than three orders of magnitude decreasedaffinity for the clinical protease inhibitors darunavir and saqui-navir, relative to the wild-type PR.[12] The crystal structure of in-hibitor-free PR20 reveals wide-open conformations of the flaps,similar to those seen in 1TW7, whereas in the inhibitor-boundPR20, the flaps were found in their usual, closed conforma-tion.[20]

We measured the 1DNH RDCs for unbound PR20 under thesame conditions as for inhibitor-free and DMP323-bound PRand used the two corresponding PR20 crystal structures, 3UCBand 3UF3, as reference structures for the closed and wide-open conformations, respectively. As for PR, the alignmenttensor parameters were determined by an SVD fit of the PR20RDCs to the core region, and these parameters were subse-quently used to predict the RDCs for the extended flap regions(Table 1). The experimental RDCs show much closer agreementwith values predicted for the wide-open PR20 conformation(Figure 3 A; Q-flap = 21.9 %) than those for the closed confor-

mation (Figure 3 B; Q-flap = 67.8 %). Therefore, in contrast to in-hibitor-free PR, the RDCs measured for PR20 clearly indicatedthat the flaps adopt a wide-open conformation in the un-bound state. The large difference between the inhibitor-freeconformation of PR and PR20 likely arises from the mutationsin the flap hinge region (E35D, M36I, and S37N), which elimi-nate several native interactions that have been postulated toimpact the orientation of the flaps.[20]

NMR relaxation data measured for inhibitor-free PR revealeddynamic fluctuations of the flap tips on the ms–ms timescale,[21] which have been commonly interpreted as reflectinga dynamic equilibrium between semi-open and wide-openconformations.[3–5] Combining the crystallographic and NMRdata with MD simulations, a popular model of the PR bindingmechanism has emerged over the past years. In this model,the inhibitor-free protease adopts a semi-open conformationwith rare and transient wide opening of the flaps that wouldallow smooth access of the natural polyproteins into the activesite. After the substrate binds to the active site, the flaps closeand then adopt the conformation observed in the protease–in-hibitor complexes. However, it is important to realize that theNMR relaxation data cannot distinguish a semi-open to wide-open conformational transition from any other mode of struc-tural fluctuation. The backbone RDCs measured in the presentstudy clearly show that the average conformation of the inhibi-tor-free PR in solution is very similar to the closed conforma-tion observed for PR–inhibitor complexes. The NMR relaxationdata reported by Freedberg et al.[21] on the inhibitor-free PRcan therefore be reinterpreted in view of the present data asreflecting transient switching between closed and a small pop-ulation of semi-open, or possibly wide-open, conformations,rather than between semi-open and wide-open conformations.Indeed, a recent MD simulation study shows a transversal pro-cession of the N-terminal cleavage site sequence (TFR/PR)through the active site without wide opening of the flaps.[6] Acomplete opening of the flaps, therefore, might not be re-quired for the catalytic mechanism of PR. In this regard, it is

Table 1. SVD-derived alignment tensor parameters obtained by a fit ofthe 1DNH RDCs to the core region of the corresponding structural model-s.[a]

Da [Hz] Rh Euler (x) Euler (y) Euler (z)

PR + DMP[b] 3BVB 2.92 0.13 43.8 �20.8 53.6PR + DMP[b] 1HHP 2.91 0.12 40.2 �17.5 56.3PR + DMP[b] 1TW7 2.84 0.15 43.3 �21.1 54.9PR apo[c] 3BVB 2.91 0.11 41.9 �19.1 52.8PR apo[c] 1HHP 2.98 0.11 39.0 �14.2 53.8PR apo[c] 1TW7 2.79 0.12 41.2 �20.0 51.5PR20 apo[d] 3UF3 3.12 0.17 40.5 �18.8 51.6PR20 apo[d] 3UCB 3.01 0.16 39.6 �14.4 46.3

[a] The core region comprises the first and last two b-sheets (residues 10–23 and 62–73), as well as the C-terminal helix (residues 87–93). The mag-nitude of the alignment tensor obtained for the apo PR and PR20 datasets was scaled up according to the 2H quadrupole splitting measured forthe PR + DMP323 sample (25.6 Hz, compared to 22 and 19 Hz for the apoPR and PR20, respectively). [b] RDC data collected for PR bound toDMP323. [c] RDC data collected for inhibitor-free PR. [d] RDC data collect-ed for inhibitor-free PR20.

Figure 3. Correlation between 1DNH RDCs measured for the unbound PR20and (A) wide-open or (B) closed structural models. RDCs corresponding tothe core region are shown in black and were used to determine the align-ment tensor through SVD fitting; flap RDCs are shown in red, with predictedvalues derived by using alignment tensor parameters obtained for the coreresidues.

ChemBioChem 2015, 16, 214 – 218 www.chembiochem.org � 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim216

Communications

also interesting to note that the flap RDCs measured in the in-hibitor-free state are not systematically reduced relative tothose seen in the Core region, indicating that dynamic averag-ing of these RDCs, which is integrated over the time scale ofmilliseconds in such RDC measurements,[22, 23] is minimal. Ourdata, therefore, paint a picture in which the flaps, with the ex-ception of their tips, have a structural rigidity that is compara-ble to the core region of the dimer and predominantly adoptthe closed conformation, even in the absence of substrate orligands. RDCs predicted for the semi-open state differ onlymodestly from those of the closed state, and our RDC datasuggest an upper limit of at most ~25 % for the semi-openstate. If the alternate state is the wide-open configuration, forwhich the predicted RDCs differ substantially more from theclosed state, an upper limit of about 10 % for the populationof this state applies.

Our results demonstrate that measurement of RDCs providesa convenient method for identifying the loop conformationsactually present in solution from a range of possibilities sug-gested by available crystal structures. This type of study isanalogous to earlier work on other flexible protein systems,such as hemoglobin[24] and lysozyme.[25] For each of these, twodifferent liquid crystalline media were used to prove that thestate of the protein is not impacted by the alignment medium.For PR, it was difficult to find liquid crystalline media compati-ble with both the free and inhibitor-bound states of the pro-tein, and we were only able to get high quality data in thenewly discovered squalamine liquid crystal. However, the factthat we can observe the wide-open state for inhibitor-freePR20, versus the closed state for inhibitor-free PR, indicatesthat the protein is not forced to adopt any particular state bythe presence of the liquid crystal.

The use of paramagnetic NMR, pseudo-contact shifts in par-ticular, provides an alternate method for probing the structureof proteins subject to dynamic rearrangement,[26] such as thestate of the HIV-1 protease flaps in solution. This method wasrecently used to investigate the structure of the flap coveringthe active site in the heterodimeric dengue virus proteaseNS2B-NS3, which was surprisingly also found to adopt theclosed conformation in the inhibitor-free state.[27] This ap-proach requires that paramagnetic lanthanide tags be linkedat suitable positions to the protein but obviates the need forfinding a suitable orienting medium needed for measurementof RDCs. In favorable cases, introduction of the paramagnetictag also yields weak protein alignment without requiring anorienting medium, thus enabling unambiguous evaluation ofthe dynamic properties of the protein.[28] Both RDC andpseudo-contact shift methods, therefore, are particularly usefulfor probing the state of dynamic regions in isotropic solution,under conditions that can closely mimic the relevant physio-logical environment.

Experimental Section

Uniformly 2H/13C/15N-enriched (>98 %) protease samples were puri-fied from inclusion bodies by size-exclusion chromatography underdenaturing conditions, followed by reversed-phase HPLC, as previ-

ously described.[29] Proteins were folded by dilution from a stocksolution of 2 mg mL�1 in HCL (12 mm) in 6.6 volumes of acetatebuffer (5 mm, pH 6.0), with or without DMP323 inhibitor, and thenwere dialyzed extensively in 20 mm sodium phosphate (pH 5.7)and concentrated. The protease constructs used in this study con-tained an additional active site D25N mutation to prevent autopro-teolysis. The D25N mutation was introduced into PR20 by Quick-Change mutagenesis.

The 1DNH RDCs were derived from the difference in 1JNH + 1DNH split-ting measured at 600 MHz using an ARTSY-HSQC experiment[30] onan isotropic sample and an aligned sample. All experiments wereperformed at 293 K. The alignment of the samples was obtainedby the addition of 10 mg mL�1 squalamine and 5 mm hexan-1-ol,yielding a stable 2H quadrupole splitting of ~20 Hz. The averageexperimental error in the measured 1DNH RDCs was 0.15 Hz. Align-ment tensor parameters are listed in Table 1.

Acknowledgements

We thank Annie Aniana, James L. Baber, and Jinfa Ying for tech-nical assistance and Dennis A. Torchia for helpful discussions. Weacknowledge support from the National Institute of Diabetes andDigestive and Kidney Diseases Advanced Mass SpectrometryCore. This work was funded by the National Institutes of Health(NIH) Intramural Research Programs of the National Institute ofDiabetes and Digestive and Kidney Diseases and by the Intramu-ral AIDS-Targeted Antiviral Program of the Office of the Director,NIH.

Keywords: active sites · flap dynamics · HIV protease · liquidcrystal NMR spectroscopy · residual dipolar coupling

[1] J. M. Louis, R. Ishima, D. A. Torchia, I. T. Weber in Advances in Pharmacol-ogy, Vol. 55: HIV-1: Molecular Biology and Pathogenesis : Viral Mecha-nisms, 2nd ed. (Ed. : K.-T. Jeang), Academic Press, San Diego, 2007,p. 261.

[2] V. Y. Torbeev, H. Raghuraman, D. Hamelberg, M. Tonelli, W. M. Westler, E.Perozo, S. B. H. Kent, Proc. Natl. Acad. Sci. USA 2011, 108, 20982.

[3] V. Hornak, A. Okur, R. C. Rizzo, C. Simmerling, Proc. Natl. Acad. Sci. USA2006, 103, 915.

[4] A. L. Perryman, J. H. Lin, J. A. McCammon, Protein Sci. 2004, 13, 1434.[5] V. Hornak, C. Simmerling, Drug Discovery Today 2007, 12, 132.[6] S. Kashif Sadiq, F. Noe, G. De Fabritiis, Proc. Natl. Acad. Sci. USA 2012,

109, 20449.[7] H. Heaslet, R. Rosenfeld, M. Giffin, Y.-C. Lin, K. Tam, B. E. Torbett, J. H.

Elder, D. E. McRee, C. D. Stout, Acta Crystallogr. Sect. D 2007, 63, 866.[8] A. H. Robbins, R. M. Coman, E. Bracho-Sanchez, M. A. Fernandez, C. T.

Gilliland, M. Li, M. Agbandje-McKenna, A. Wlodawer, B. M. Dunn, R.McKenna, Acta Crystallogr. Sect. D 2010, 66, 233.

[9] J. H. Prestegard, C. M. Bougault, A. I. Kishore, Chem. Rev. 2004, 104,3519.

[10] A. Bax, A. Grishaev, Curr. Opin. Struct. Biol. 2005, 15, 563.[11] P. Y. S. Lam, P. K. Jadhav, C. J. Eyermann, C. N. Hodge, Y. Ru, L. T. Bacheler,

J. L. Meek, M. J. Otto, M. M. Rayner, Y. N. Wong, C. H. Chang, P. C. Weber,D. A. Jackson, T. R. Sharpe, S. Ericksonviitanen, Science 1994, 263, 380.

[12] J. M. Louis, A. Aniana, I. T. Weber, J. M. Sayer, Proc. Natl. Acad. Sci. USA2011, 108, 9072.

[13] A. S. Maltsev, A. Grishaev, J. Roche, M. Zasloff, A. Bax, J. Am. Chem. Soc.2014, 136, 3752.

[14] J. M. Sayer, F. Liu, R. Ishima, I. T. Weber, J. M. Louis, J. Biol. Chem. 2008,283, 13459.

[15] S. Spinelli, Q. Z. Liu, P. M. Alzari, P. H. Hirel, R. J. Poljak, Biochimie 1991,73, 1391.

ChemBioChem 2015, 16, 214 – 218 www.chembiochem.org � 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim217

Communications

[16] P. Martin, J. F. Vickrey, G. Proteasa, Y. L. Jimenez, Z. Wawrzak, M. A. Win-ters, T. C. Merigan, L. C. Kovari, Structure 2005, 13, 1887.

[17] G. Cornilescu, J. L. Marquardt, M. Ottiger, A. Bax, J. Am. Chem. Soc.1998, 120, 6836.

[18] A. Bax, Protein Sci. 2003, 12, 1.[19] I. Dierynck, M. De Wit, E. Gustin, I. Keuleers, J. Vandersmissen, S. Hallen-

berger, K. Hertogs, J. Virol. 2007, 81, 13845.[20] J. Agniswamy, C.-H. Shen, A. Aniana, J. M. Sayer, J. M. Louis, I. T. Weber,

Biochemistry 2012, 51, 2819.[21] D. I. Freedberg, R. Ishima, J. Jacob, Y.-X. Wang, I. Kustanovich, J. M.

Louis, D. A. Torchia, Protein Sci. 2002, 11, 221.[22] J. R. Tolman, J. Am. Chem. Soc. 2002, 124, 12020.[23] O. F. Lange, N. A. Lakomek, C. Fares, G. F. Schroder, K. F. A. Walter, S.

Becker, J. Meiler, H. Grubm�ller, C. Griesinger, B. L. de Groot, Science2008, 320, 1471.

[24] J. A. Lukin, G. Kontaxis, V. Simplaceanu, Y. Yuan, A. Bax, C. Ho, Proc. Natl.Acad. Sci. USA 2003, 100, 517.

[25] N. K. Goto, N. R. Skrynnikov, F. W. Dahlquist, L. E. Kay, J. Mol. Biol. 2001,308, 745.

[26] L. Cerofolini, G. B. Fields, M. Fragai, C. F. G. C. Geraldes, C. Luchinat, G.Parigi, E. Ravera, D. I. Svergun, J. M. C. Teixeira, J. Biol. Chem. 2013, 288,30659.

[27] L. de La Cruz, H. D. N. Thi, K. Ozawa, J. Shin, B. Graham, T. Huber, G.Otting, J. Am. Chem. Soc. 2011, 133, 19205.

[28] I. Bertini, C. Del Bianco, I. Gelis, N. Katsaros, C. Luchinat, G. Parigi, M.Peana, A. Provenzani, M. A. Zoroddu, Proc. Natl. Acad. Sci. USA 2004,101, 6841.

[29] R. Ishima, D. A. Torchia, J. M. Louis, J. Biol. Chem. 2007, 282, 17190.[30] N. C. Fitzkee, A. Bax, J. Biomol. NMR 2010, 48, 65.

Received: October 8, 2014

Published online on December 2, 2014

ChemBioChem 2015, 16, 214 – 218 www.chembiochem.org � 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim218

Communications